Imaging the Intersection of Parenchymal Abnormalities and Pulmonary Vascular Pathways

In patients with pulmonary hypertension (PH), mild elevations in pulmonary arterial pressure and resistance have been shown to significantly associate with increased mortality (1). In the setting of chronic lung disease, PH is a common complication that significantly associates with mortality but to varying degrees, depending on the underlying lung disease, including chronic obstructive pulmonary disease (COPD), idiopathic pulmonary fibrosis, and interstitial lung disease (2). The pulmonary vascular bed is normally a highly compliant system that requires at least 50% of the pulmonary vasculature to be obstructed to appreciate an increase in pulmonary arterial pressure (3). So, although there is a growing appreciation for pulmonary vasculopathy beginning early in lung diseases, there need to be significant vascular abnormalities for pulmonary pressure to reach current hemodynamic thresholds for PH (4). Patients with early signs of interstitial abnormalities already exhibit a decrease in exercise capacity and an increase in shortness of breath (5). Noninvasive computed tomography (CT) provides an opportunity to quantitatively assess the parenchymal and pulmonary vascular structures at the same time.

The group at Brigham and Women’s Hospital have pioneered the computational and analytical methods to quantitatively assess interstitial lung abnormalities, pulmonary arterial vasculature, and the right and left ventricular volumes from thin-slice chest CT (6–8). In their chest imaging platform, they developed and trained an automated method to detect radiographic features in CT images that associate with interstitial subtypes (reticulations, honeycombing, centrilobular nodules, linear scar, nodular changes, subpleural line, and ground glass) or emphysema subtypes (centrilobular, paraseptal) (Figure 1A). The proportion of interstitial features is termed the “quantitative interstitial abnormalities” (QIA). Using the same images, they can reconstruct the pulmonary vasculature to quantify clinically meaningful cardiopulmonary features, including blood volume in the preacinar arteries (aBV5-20/TBV), smaller distal vessels (aBV5/TBV), and ventricular volumes (Figure 1A). Comparisons of the blood volumes in smaller arteries with cross-sectional areas <5 mm² to intermediate-sized vessels >10 mm² is becoming more protocolized to characterize clinical vascular remodeling from CT imaging (9).

Figure 1.

(A) Depiction of the pulmonary vasculature and corresponding airways. The preacinar arteries are more intermediate-sized arteries (aBV5–20 mm²) that are smaller than the main pulmonary artery but larger than the more distal vasculature (14). The CT/radiographic features are used to quantify the percentage of the parenchyma that is associated with QIA and percentage emphysema (6). (B) Highlighting the progressive impact of the pulmonary vascular dysfunction on the mediation of progression of the casual mediation results of the metrics of pulmonary vascular dysfunction. The metrics that measure more proximal features, including the RV/LV ratio and the PA/Ao ratio, were suppressive and partial mediators of the clinical outcome of the 6MWD, respectively. The main mediator was the preacinar arteries that mediated 79.6% of the outcome. 6MWD = 6-minute-walk distance; CT = computed tomography; PA/Ao ratio = pulmonary to aorta ratio; QIA = quantitative interstitial abnormalities; RV/LV ratio = right to left ventricle ratio.

In this issue of the Journal, Harder and colleagues (pp. 1132–1142) used causal mediation analysis to determine that 6-minute-walk distance (6MWD) and the modified Medical Research Council dyspnea scale score are partially mediated by dilation in the preacinar arteries in ever-smokers from the Genetic Epidemiology of COPD (COPDGene) study cohort (n = 8,200) (10). To do this, they used radiographic features of QIA and three CT-based vascular measures and right heart metrics, including the ratio of the right/left ventricular volumes (RV/LV ratio), the pulmonary artery to aorta ratio (PA/Ao ratio), and the preacinar arterial blood volume (aBV5-20/TBV, PA volume 5–20 mm²/total arterial volume cross-sectional area) (Figure 1B). The cohort had a median QIA burden of 4.67 (2.99–7.48%) with a median 6MWD of 425 [343–496] meters. They found that the QIA percentage correlated with decreased exercise capacity where a 1% increase in QIA was associated with a 2-meter decline in 6MWD. QIA was associated with all three of the pulmonary vascular and right heart metrics. but it was the preacinar arterial blood volume that partially mediated (79.6%) the effect of QIA burden on the clinical outcomes (Figure 1B). Interestingly, the PA/Ao ratio was a weak mediator, and RV/LV was a suppressor. Using proteomic analysis in a subset of the cohort, they found that preacinar arterial dilation was also associated with increased protein levels of known pulmonary vascular dysfunction biomarkers, including angiopoietin-2 and N-terminal prohormone of brain natriuretic peptide. These results suggest that the noninvasive CT measures can be used to quantify some of the burden of disease before significant parenchymal damage.

A strength of this study is the integration of radiometric analysis, automated segmentation methods, and causal mediation analysis to identify potential noninvasive biomarkers of early pulmonary vascular dysfunction before the onset of overt lung disease or pulmonary vascular disease. It is a tour de force to perform this comprehensive analysis on 8,200 CT scans. This analysis demonstrates the wealth of information in a CT scan and the potential expanded role for CT imaging in the diagnosis and treatment of patients with unexplained dyspnea and decreased exercise tolerance. A significant limitation of the present analysis is that these findings have not been correlated with right heart catheterization–measured pulmonary arterial pressure or pulmonary vascular resistance, because these tests were not included in the COPDGene protocol. Despite not knowing the pulmonary pressure, it makes sense that the RV/LV ratio and PA/Ao ratio would not mediate functional outcomes on the basis of their distance from the primary parenchymal site of disease and the logical anatomical progression of remodeling. The same group had previously demonstrated preacinar arterial dilation in patients with idiopathic interstitial pneumonia (11), and their CT-based features correlate with histological measures of pulmonary vascular remodeling (12). Studies have shown dilation of the more proximal vasculature with increased pulmonary vascular resistance due to the elevated pulmonary pressure (13). The present results demonstrate that preacinar arterial dilation might be an early biomarker that mediates the decrease in 6MWD and increased dyspnea. It is not surprising that the more proximal measures of pulmonary vascular dysfunction, including the PA/Ao ratio and the RV/LV ratio, were not significant mediators of the outcomes in this cohort with interstitial lung abnormalities and presumable minimal pulmonary vascular disease. PA diameter significantly correlates with mean PA pressure and is important in the screening of suspected PH (13). Because of the limitations in the available data in COPDGene, this study is unable to provide a definitive association between pulmonary vascular resistance and pressure with the CT-based vascular features. Further validation of this approach in a cohort with CT imaging and pulmonary artery pressure would be warranted.

Despite these limitations, Harder and colleagues present an intricate study that contributes to our understanding of the clinical meaning of these noninvasive radiometric features of the parenchyma and pulmonary vasculature. Automated methodologies in this approach make these measures more accessible in clinical settings in patients with lung abnormalities and early forms of lung disease. This quantitative and comprehensive analysis of CT images expands our capabilities for identifying functional disease biomarkers in early disease.